Hugging Face's logo

Datasets:
optimum-benchmark
/
cpu

Dataset card Viewer Files Community
IlyasMoutawwakil HF staff commited on
Commit
a4239be
•
1 Parent(s): 17bdd8f

Upload cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json with huggingface_hub

Browse files
Files changed (1) hide show
  1. cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json +80 -74
cpu_inference_transformers_text-generation_openai-community/gpt2/benchmark.json CHANGED
@@ -73,14 +73,14 @@
73
  "environment": {
74
  "cpu": " AMD EPYC 7763 64-Core Processor",
75
  "cpu_count": 4,
76
- "cpu_ram_mb": 16757.346304,
77
  "system": "Linux",
78
  "machine": "x86_64",
79
  "platform": "Linux-6.5.0-1018-azure-x86_64-with-glibc2.35",
80
  "processor": "x86_64",
81
  "python_version": "3.10.14",
82
  "optimum_benchmark_version": "0.2.0",
83
- "optimum_benchmark_commit": "b0a0b933f18d5fa56eadb290a26cc2099f2dd4d8",
84
  "transformers_version": "4.40.2",
85
  "transformers_commit": null,
86
  "accelerate_version": "0.30.1",
@@ -99,7 +99,7 @@
99
  "prefill": {
100
  "memory": {
101
  "unit": "MB",
102
- "max_ram": 947.67104,
103
  "max_global_vram": null,
104
  "max_process_vram": null,
105
  "max_reserved": null,
@@ -107,49 +107,51 @@
107
  },
108
  "latency": {
109
  "unit": "s",
110
- "count": 12,
111
- "total": 0.6362601239999606,
112
- "mean": 0.053021676999996714,
113
- "stdev": 0.0009879796796853479,
114
- "p50": 0.05330665149999447,
115
- "p90": 0.054035152600016546,
116
- "p95": 0.054086199050016145,
117
- "p99": 0.054134867010011474,
118
  "values": [
119
- 0.05414703400001031,
120
- 0.05403642500002093,
121
- 0.05157386299998734,
122
- 0.05316608200001838,
123
- 0.05345428299997934,
124
- 0.05070891999997684,
125
- 0.053372851000005994,
126
- 0.05278962599999204,
127
- 0.053284173999998075,
128
- 0.05237403600000334,
129
- 0.05402370099997711,
130
- 0.05332912899999087
 
 
131
  ]
132
  },
133
  "throughput": {
134
  "unit": "tokens/s",
135
- "value": 37.72042140425177
136
  },
137
  "energy": {
138
  "unit": "kWh",
139
- "cpu": 1.8332425392035284e-06,
140
- "ram": 7.661293786705932e-08,
141
  "gpu": 0.0,
142
- "total": 1.909855477070588e-06
143
  },
144
  "efficiency": {
145
  "unit": "tokens/kWh",
146
- "value": 1047199.6567340684
147
  }
148
  },
149
  "decode": {
150
  "memory": {
151
  "unit": "MB",
152
- "max_ram": 947.67104,
153
  "max_global_vram": null,
154
  "max_process_vram": null,
155
  "max_reserved": null,
@@ -157,74 +159,78 @@
157
  },
158
  "latency": {
159
  "unit": "s",
160
- "count": 12,
161
- "total": 0.39697931500001005,
162
- "mean": 0.03308160958333417,
163
- "stdev": 0.0006865410009691628,
164
- "p50": 0.03313834499999757,
165
- "p90": 0.03381396359999087,
166
- "p95": 0.034019586199990216,
167
- "p99": 0.034196957239990694,
168
  "values": [
169
- 0.032926583000005394,
170
- 0.03359598000000119,
171
- 0.03303443499999048,
172
- 0.03256758899999568,
173
- 0.03232157700000471,
174
- 0.03324225500000466,
175
- 0.03343429599999581,
176
- 0.0327964890000203,
177
- 0.034241299999990815,
178
- 0.033838183999989724,
179
- 0.033405552000004946,
180
- 0.031575075000006336
 
 
181
  ]
182
  },
183
  "throughput": {
184
  "unit": "tokens/s",
185
- "value": 30.228275244012895
186
  },
187
  "energy": {
188
  "unit": "kWh",
189
- "cpu": 1.0896824827097892e-06,
190
- "ram": 4.553964979426669e-08,
191
  "gpu": 0.0,
192
- "total": 1.1352221325040545e-06
193
  },
194
  "efficiency": {
195
  "unit": "tokens/kWh",
196
- "value": 880884.8694609365
197
  }
198
  },
199
  "per_token": {
200
  "memory": null,
201
  "latency": {
202
  "unit": "s",
203
- "count": 11,
204
- "total": 0.9463848640000094,
205
- "mean": 0.0860349876363645,
206
- "stdev": 0.0011781546398233853,
207
- "p50": 0.08581834000000299,
208
- "p90": 0.08754055299999663,
209
- "p95": 0.08759233499999652,
210
- "p99": 0.08763376059999643,
211
  "values": [
212
- 0.08764411699999641,
213
- 0.08462633300001698,
214
- 0.08574536300000091,
215
- 0.08581834000000299,
216
- 0.08396824700000138,
217
- 0.08685305199998083,
218
- 0.08559130400001891,
219
- 0.08754055299999663,
220
- 0.08622163699999419,
221
- 0.08742857199999321,
222
- 0.08494734600000697
 
 
223
  ]
224
  },
225
  "throughput": {
226
  "unit": "tokens/s",
227
- "value": 11.623178284474223
228
  },
229
  "energy": null,
230
  "efficiency": null
 
73
  "environment": {
74
  "cpu": " AMD EPYC 7763 64-Core Processor",
75
  "cpu_count": 4,
76
+ "cpu_ram_mb": 16757.338112,
77
  "system": "Linux",
78
  "machine": "x86_64",
79
  "platform": "Linux-6.5.0-1018-azure-x86_64-with-glibc2.35",
80
  "processor": "x86_64",
81
  "python_version": "3.10.14",
82
  "optimum_benchmark_version": "0.2.0",
83
+ "optimum_benchmark_commit": "3e2eebdc0f80ae6deeb2e1faad3e889ed5a4df2d",
84
  "transformers_version": "4.40.2",
85
  "transformers_commit": null,
86
  "accelerate_version": "0.30.1",
 
99
  "prefill": {
100
  "memory": {
101
  "unit": "MB",
102
+ "max_ram": 958.312448,
103
  "max_global_vram": null,
104
  "max_process_vram": null,
105
  "max_reserved": null,
 
107
  },
108
  "latency": {
109
  "unit": "s",
110
+ "count": 14,
111
+ "total": 0.6226214250000339,
112
+ "mean": 0.04447295892857385,
113
+ "stdev": 0.0006497683664670401,
114
+ "p50": 0.04445699649998858,
115
+ "p90": 0.04529145000000767,
116
+ "p95": 0.04560630480000612,
117
+ "p99": 0.045885101759996015,
118
  "values": [
119
+ 0.045954800999993495,
120
+ 0.04446333299998173,
121
+ 0.044494441000011875,
122
+ 0.04426547500000311,
123
+ 0.04376506200000563,
124
+ 0.04499464299999545,
125
+ 0.04426369099999761,
126
+ 0.04452789300000859,
127
+ 0.043635319999992817,
128
+ 0.04387021799999502,
129
+ 0.04445065999999542,
130
+ 0.04541865300001291,
131
+ 0.043626325000019506,
132
+ 0.04489091000002077
133
  ]
134
  },
135
  "throughput": {
136
  "unit": "tokens/s",
137
+ "value": 44.97114759582595
138
  },
139
  "energy": {
140
  "unit": "kWh",
141
+ "cpu": 1.6963920155585101e-06,
142
+ "ram": 7.089398842205409e-08,
143
  "gpu": 0.0,
144
+ "total": 1.767286003980564e-06
145
  },
146
  "efficiency": {
147
  "unit": "tokens/kWh",
148
+ "value": 1131678.741016044
149
  }
150
  },
151
  "decode": {
152
  "memory": {
153
  "unit": "MB",
154
+ "max_ram": 958.312448,
155
  "max_global_vram": null,
156
  "max_process_vram": null,
157
  "max_reserved": null,
 
159
  },
160
  "latency": {
161
  "unit": "s",
162
+ "count": 14,
163
+ "total": 0.3892796470000519,
164
+ "mean": 0.027805689071432278,
165
+ "stdev": 0.0009139966924691778,
166
+ "p50": 0.02751088150000669,
167
+ "p90": 0.028758083700014935,
168
+ "p95": 0.0293590112500155,
169
+ "p99": 0.030069515850009054,
170
  "values": [
171
+ 0.0284717760000035,
172
+ 0.027489577000011423,
173
+ 0.028459322999992764,
174
+ 0.027429003000008834,
175
+ 0.02726347500001225,
176
+ 0.027557150999996338,
177
+ 0.027213382000013553,
178
+ 0.02688716399998725,
179
+ 0.027532186000001957,
180
+ 0.028088781999997536,
181
+ 0.028880787000019836,
182
+ 0.02703322499999672,
183
+ 0.030247142000007443,
184
+ 0.026726674000002504
185
  ]
186
  },
187
  "throughput": {
188
  "unit": "tokens/s",
189
+ "value": 35.963863273843685
190
  },
191
  "energy": {
192
  "unit": "kWh",
193
+ "cpu": 1.0036530932366561e-06,
194
+ "ram": 4.1946113965940766e-08,
195
  "gpu": 0.0,
196
+ "total": 1.0455992072025971e-06
197
  },
198
  "efficiency": {
199
  "unit": "tokens/kWh",
200
+ "value": 956389.4015139955
201
  }
202
  },
203
  "per_token": {
204
  "memory": null,
205
  "latency": {
206
  "unit": "s",
207
+ "count": 13,
208
+ "total": 0.9377388369999835,
209
+ "mean": 0.07213375669230643,
210
+ "stdev": 0.0007910113430319274,
211
+ "p50": 0.07195642499999622,
212
+ "p90": 0.0731509747999894,
213
+ "p95": 0.07340619199998741,
214
+ "p99": 0.07365363199999024,
215
  "values": [
216
+ 0.07198634099998458,
217
+ 0.07295490600000676,
218
+ 0.07172076599999855,
219
+ 0.07105912399998715,
220
+ 0.07256558100002053,
221
+ 0.07149065799998766,
222
+ 0.07143760900001439,
223
+ 0.07115158700000279,
224
+ 0.07195642499999622,
225
+ 0.07319999199998506,
226
+ 0.07267799000001673,
227
+ 0.07371549199999095,
228
+ 0.07182236599999214
229
  ]
230
  },
231
  "throughput": {
232
  "unit": "tokens/s",
233
+ "value": 13.863134901812996
234
  },
235
  "energy": null,
236
  "efficiency": null